The invention relates to 2-dimensional capacitive position sensors, typically actuated by a human finger, or a stylus. Example devices include touch screens and touch pads, particularly those over liquid crystal displays (LCDs), or cathode ray tubes (CRTs) and other types of displays, or pen-input tablets, or encoders used in machinery for feedback control purposes. Descriptions of pen or touch input to a machine date back to at least 1908, as embodied in patent DE 203,719.
Touch screens and pointing devices have become increasingly popular and common not only in conjunction with personal computers but also in all manner of other appliances such as personal digital assistants (PDAs), point of sale (POS) terminals, electronic information and ticketing kiosks, kitchen appliances and the like. These devices are evolving continuously into lower priced products and as a result, there is a need for ever lower production cost while maintaining high levels of quality and robustness.
Touch screens generally fall into two categories, capacitive and resistive.
For capacitive devices, the term ‘two-dimensional capacitive transducer’ or ‘2DCT’ is used as a generic term for touch screens, touch sensing pads, proximity sensing areas, display overlay touch screens over LCD, plasma, or CRT screens or the like, position sensing for mechanical devices or feedback systems, or other types of control surfaces without limitation, having a surface or volume capable of reporting at least a 2-dimensional coordinate, Cartesian or otherwise, related to the location of an object or human body part, by means of a capacitance sensing mechanism.
For resistive devices, the term ‘two-dimensional resistive transducer’ or ‘2DRT’ is used as a generic term to refer to touch screens or pen input devices based on purely galvanic principles.
The term ‘2DxT’ refers to elements of either the 2DCT or 2DRT type.
The term ‘touch’ means touch or proximity by a human body part or mechanical component of sufficient capacitive signal strength to generate a desired output. In the sense of ‘proximity’, touch can also mean to ‘point’ at a 2DCT without making physical contact, where the 2DCT responds to the capacitance from the proximity of the object sufficient to react properly.
The term ‘element’ refers to the active sensing element of a 2DCT or 2DRT. The term ‘electrode’ refers to a connection point at the periphery of the element.
The term ‘stripe’ refers to an electrical line conductor that is a component part of an element and which has two ends. A stripe can be a wire. A stripe can have substantial galvanic resistance by intent, whereas a wire has minimal resistance. If the element of which it is a part is physically curved, the stripe would also be physically curved.
The term ‘pin cushion’ refers to any distortion of the signal from a 2DCT whether parabolic, barrel, or other form of 2D dimensional aberration.
Many types of 2DCT are known to suffer from geometric distortion characterized as ‘pin cushion’ or ‘hyperbolic’ or ‘parabolic’, whereby the reported coordinate of touch is in error due to electrical effects on the sensing surface. These effects are described in more depth in various other patents, for example in Pepper U.S. Pat. No. 4,198,539, incorporated by reference. An excellent summary of the known causes, solutions, and problems of the solutions to geometric distortion can be found in a reading of Babb et al, in U.S. Pat. No. 5,940,065 and U.S. Pat. No. 6,506,983, incorporated by reference. U.S. Pat. No. 5,940,065 describes succinctly the two major classes of correction: 1) Electromechanical methods involving design of or modification to the sensing surface or the connecting electrodes; 2) Modeling methods using mathematical algorithms to correct the distortions.
Electromechanical Methods
Edge Manipulation of Planar Element: Küpfmüller et al in U.S. Pat. No. 2,338,949 (filed 1940) solve the problem of edge distortion in a 2DRT electrograph using very long rectangular tails in X and Y surrounding a small usable area. Küpfmüller takes the further approach of slotting the four tails into stripes; these stripes do not intrude on the user input area but do act to raise the resistance to current flow in an anisotropic manner along sides parallel to current flow. This idea reappears in slightly different form in Yaniv et al, U.S. Pat. No. 4,827,084, nearly 50 years later. Küpfmüller remains the most similar prior art to the instant invention.
Becker in U.S. Pat. No. 2,925,467 appears the first to describe a 2DRT electrograph whereby nonlinear edge effects are eliminated via the use of a very low resistance edge material relative to the sheet resistance of the element proper. This method can also be used to construct a 2DCT.
Pepper, in U.S. Pat. No. 4,198,539, U.S. Pat. No. 4,293,734, and U.S. Pat. No. 4,371,746 describes methods of linearizing a 2DCT by manipulating the edge resistance structure of the element.
Talmage, in U.S. Pat. No. 4,822,957 describes a similar edge pattern as Pepper in conjunction with a 2DRT element and a pick-off sheet. Numerous other such patents have been issued using various methods, and the area remains a fertile one for new patents to this day. These methods have been found to be very difficult to develop and replicate, and they are prone to differential thermal heating induced errors and production problems. Very small amounts of localized error or drift can cause substantial changes in coordinate response. The low resistance of the patterned edge strips causes problems with the driver circuitry, forcing the driver circuitry to consume more power and be much more expensive than otherwise. There are a significant number of patents that reference the Pepper patents and which purport to do similar things. The improvements delivered by Pepper etc over Becker are arguably marginal, as at least Becker is easier and more repeatable to fabricate.
Edge Resistance with Wire Element: Kable in U.S. Pat. No. 4,678,869 discloses a 2D array for pen input, using resistive divider chains on 2 axes with highly conductive electrodes connected to the chains, the electrodes having some unintended resistance for the purposes of detection, and the detection signal being interpolated from the signals generated between two adjacent electrodes. The unintended resistance causes a slight amount of pin cushion in the response. This patent also describes an algorithmic means to compensate for the slight pin-cushion distortion developed by this technique. The Kable method is not operable with other than a connected stylus, i.e. it is not described as being responsive to a human finger. The Kable patent requires crossovers between conductors and thus needs at least three construction layers (conductor, insulator, conductor).
Multiple Active-Edge Electrodes: Turner in U.S. Pat. No. 3,699,439 discloses a uniform resistive screen with an active probe having multiple electrode connections on all four sides to linearize the result.
Yoshikawa et al, in U.S. Pat. No. 4,680,430, and Wolfe, in U.S. Pat. No. 5,438,168, teach 2DCT's using multiple electrode points on each side (as opposed to the corners) to facilitate a reduction in pin cushion by reducing the interaction of the current flow from the electrodes on one axis with the electrodes of the other. While the element is a simple sheet resistor, this approach involves large numbers of active electronic connections (such as linear arrays of diodes or MOSFETs) at each connection point in very close proximity to the element.
Nakamura in U.S. Pat. No. 4,649,232 teaches similarly as Yoshikawa and Wolfe but with a resistive pickup stylus.
Sequentially Scanned Stripe Element: Greanias et al in U.S. Pat. No. 4,686,332 and U.S. Pat. No. 5,149,919, Boie et al in U.S. Pat. No. 5,463,388, and Landmeier in U.S. Pat. No. 5,381,160 teach methods of element sensing using alternating independently driven and sensed stripe conductors in both the X and Y axis, from which is interpreted a position of a finger touch or, by a pickup device, a stylus pen. The construction involves multiple layers of material and special processing. Greanias teaches the use of interpolation between the stripes to achieve higher resolution in both axis. Both require three or more layers to allow crossovers of conductors within the element. Both rely on measurements of capacitance on each stripe, not the amount of cross coupling from one stripe to another. Boie also teaches a special guard-plane.
Binstead, in U.S. Pat. No. 5,844,506 and U.S. Pat. No. 6,137,427 teaches a touch screen using discrete fine wires in a manner similar to those taught by Kable, Allen, Gerpheide and Greanias. Binstead uses very fine row and column wires to achieve transparency. This patent also teaches the Greanias method of interpolation between electrode wires to achieve higher resolution. The scanning relies on measurements of capacitance on each stripe to ground, not the amount of cross coupling from one to another.
Evans in U.S. Pat. No. 4,733,222 also describes a system wherein stripes are sequentially driven in both X and Y axis, using also an external array of capacitors to derive sensing signals via a capacitor divider effect. Interpolation is used to evaluate finer resolutions than possible with the stripes alone.
Volpe in U.S. Pat. No. 3,921,166 describes a discrete key mechanical keyboard that uses a capacitive scanning method. There are sequentially driven input rows and sequentially sensed columns. The press of a key increases the coupling from a row to a column, and in this way n-key rollover can be achieved; there is no need for interpolation. Although not a 2DCT, Volpe presages scanned stripe element 2DCT technology. My own U.S. Pat. No. 6,452,514 also falls into this classification of sensor.
Itaya in U.S. Pat. No. 5,181,030 discloses a 2DRT having resistive stripes which couple under pressure to a resistive plane which reads out the location of contact. The stripes, or the plane, have a 1D voltage gradient imposed on them so that the location of contact on particular the stripe can be readily identified. Each stripe requires its own, at least one electrode connection.
Cyclical Scanned Stripe Element: Gerpheide et al, in U.S. Pat. No. 5,305,017 teaches a touch-pad capacitance-based computer pointing device using multiple orthogonal arrays of overlapping metallic stripes separated by insulators. The scan lines are arranged in a cyclically repeating pattern to minimize drive circuitry requirements. A cyclical nature of the wiring of the invention prevents use of this type of 2DCT for absolute position location. The invention is suited to touch pads used to replace mice, where actual location determination is not required, and only relative motion sensing is important. Gerpheide teaches a method of signal balance between two phase-opposed signals at the location of touch.
Parallel Read Stripe Elements: Allen et al in U.S. Pat. No. 5,914,465 teach an element having rows and column scan stripes which are read in parallel by analog circuitry. The patent claims lower noise and faster response times than sequentially scanned elements. The method is particularly suited to touch pads for mouse replacement but does not scale well to higher sizes. Multiple construction layers are required as with all stripe element 2DCT's. The Allen method requires large scale integration and high numbers of connection pins. It interpolates to achieve higher resolution than achievable by the number of raw stripes.
In WO 04/040240, “Charge Transfer Capacitive Position Sensor”, Philipp describes in conjunction with FIG. 12 a method of using individual resistive 1-D stripes to create a touch screen. These stripes can be read either in parallel or sequentially, since the connections to these stripes are independent of one another. Furthermore, in connection with FIG. 6 there is described an interpolated coupling between adjacent lumped electrode elements and an object such as a finger. WO 04/040240 is incorporated herein by reference.
In WO 2005/020056 Philipp describes a position sensor having first and second resistive bus-bars spaced apart with an anisotropic conductive area between them (see FIG. 3 thereof). Electric currents induced in the anisotropic conductive area by touch or proximity flow preferentially towards the bus-bars to be sensed by detection circuitry. Because induced currents, for example those induced by drive circuitry, flow preferentially along one direction, pin-cushion distortions in position estimates are largely constrained to this single direction. Such one-dimensional distortions can be corrected for very simply by applying scalar correction factors, thereby avoiding the need for complicated vector correction. This provides a 2D pattern of conductive material for sensing capacitance behind a plastic or glass panel or other dielectric, which can be used as a 2DxT, whether in the format of a touch screen or ‘touch pad’. The conductor can be clear, for example made of indium tin oxide (ITO), to provide a suitable transparent overlay for a display or other backing.
This approach works well for relatively small screen sizes up to about a 2 inch (50 mm) diagonal suitable for a cellular telephone but performance degrades with larger screens such as needed for some white goods, for example a microwave oven. Moreover, the handshadow effect can cause problems with this design.
In U.S. Pat. No. 6,288,707 Philipp describes a capacitive position sensor intended to function as part of a computer pointing device that employs ratiometric capacitive sensing techniques. An array of patterned metallic electrodes is disposed on an insulating substrate layer, where the electrode geometry is selected to generate a varying capacitive output as a user's finger moves across the electrode array.
FIG. 1 of the accompanying drawings reproduces FIG. 4 of U.S. Pat. No. 6,288,707. An array of patterned metallic electrodes is disposed on an insulating layer where the electrode geometry is selected to generate a varying capacitive output as a user's finger moves across the electrode array. This arrangement comprises four interspersed electrode sets, two for each dimension. The x-axis sets, which are triangular, are easier to see and understand. A first set of triangles 1 are all electrically connected together to an output bus denoted as X1. The second set 2 are also connected together to an output labeled X2. The position of a user's hand with respect to the x-axis can be ascertained from the ratio of signals from X1 and X2. Because capacitance is directly proportional to surface area, and because the plates connected to X1 aggregate to a greater surface area to the left than do the plates connected to X2 (and vice versa) the ability to take the ratio of X1/X2 or X2/X1 is preserved so long as a great enough finger area is over the pattern at a close enough range to provide sufficient signal strength. A corresponding set of plates are connected to the Y1 and Y2 buses. The Y-connected set is also ratiometric, although in a manner different from the X sets. The Y set consists of alternating Y1-connected and Y2-connected rectangular stripes, 3 and 4 respectively, having a y-axis dimension that varies with placement in such a manner so as to create a smoothly varying ratio of surface area between Y1 and Y2 with location Y. The sum of each adjacent pair of the y-axis stripes 3 and 4 is made constant so that the sum of the capacitance is the same for any two paired stripes, i.e., C(Y1)+C(Y2)=C(Y) for each pair of stripes. Then, as the user's fingers move along the y-axis, the detected capacitance ratio is measured in the same manner as the CX1)/C(X2) ratio, i.e. the largest value becomes the numerator. However, this design offers limited capability for the 2DCT dimension in the x-direction.
Numerical Methods
Nakamura in U.S. Pat. No. 4,650,926 describes a system for numerical correction of an electrographic system such as a tablet, using a lookup table system to correct raw 2D coordinate data.
Drum, in U.S. Pat. No. 5,101,081 describes a system for numerical correction of an electrographic system such as a tablet via remote means.
McDermott in U.S. Pat. No. 5,157,227 teaches a numerical method of correcting a 2DxT employing stored constants which are used during operation to control one or more polynomials to correct the location of reported touch by zone or quadrant.
Babb et al, in U.S. Pat. Nos. 5,940,065 and 6,506,983 teach a numerical method to linearize a 2DxT uniform sheet element using coefficients determined during a learn process, without segmentation by zone or quadrant, and on an individual unit basis so as to correct for even minor process variations. The methods disclosed by Babb are complex and involve ‘80 coefficients’ and fourth order polynomials, the coefficients of which must be determined through a rigorous and time-consuming calibration procedure. In tests supervised by the instant inventor, it has been found that 6th order polynomials are required to produce accuracy levels that are acceptable in normal use, and that the result is still highly prone to the slightest subsequent variations post-calibration due to thermal drift and the like. In particular it has been found that the corner connections are extreme contributors to long-term coordinate fluctuations, as they act as singularities with a high gain factor with respect to connection size and quality. Furthermore, the method of numerical correction requires high-resolution digital conversions in order to produce even modest resolution outputs. For example it has been found that a 14-bit ADC is required to provide a quality 9-bit coordinate result. The extra expense and power required of the amplifier system and ADC can be prohibitive in many applications.
Problem
Despite the extensive prior work in this field, there is still a need for a low-cost, single-layer, large-area, transparent, low-distortion 2DCT with a relatively low number of external connections.